No. 593] THE MECHANISM OF CROSSING-OVER 



297 



this, nor fell again later, and if cytological measurements 

 should then substantiate the judgment, based on inspec- 

 tion, that the loops did have a modal length during the 

 strepsinema stage, there would be good evidence that 

 crossing-over must occur at an early stage of synapsis. 



Other peculiarities of coincidence also might be found 

 which would permit of explanation on one scheme and 

 not on another. In groups II and III, for example, there 

 seem to be peculiarities in the coincidence relations in 

 cases where the chromosomes differ in regard to the fac- 

 tor C, or a similar factor. And a comparison of coinci- 

 dence in different regions of the chromosome in any given 

 case or in the same region of the chromosome in cases of 

 linkage variation, might very well reveal relations that 

 lend evidence to one scheme of crossing-over or another. 

 Even a determination, not of coincidence, but merely of 

 linkage variation itself, in different parts of the chromo- 

 some, might in some way shed light on the subject. In 

 the case of the third chromosome, experiments of this sort 

 are now under way with multiple stocks which I have 

 made up for this purpose, and Sturtevant is conducting 

 similar experiments with group II. The first require- 

 ment, however, is obviously an accurate study of normal 

 coincidence, and it therefore became necessary to deter- 

 mine the coincidence for points various distances apart, 

 preferably in the same experiment. But to work with a 

 great many factors in a group at once introduced new diffi- 

 culties, which made special methods necessary, as will be 

 explained later. Before considering this experiment, it 

 will be desirable to consider other lines of evidence and 

 modes of attacking the problem of crossing-over. 



The cytological evidence which Janssens presents for 

 crossing-over is entirely directed towards proving that 

 crossing-over occurs during strepsinema or later. In 

 strepsinema the chromosomes, as already mentioned, 



(at J), it is easily seen that the proportion of breaks at K would be lower 

 when breakage oeeure.l at I than when there was no breakage at I, whereas 



higher when there was breakage at I than when there was no breakage at I. 



